Inc.) 2009-2012 JSPS DC1 2012-2014 Research Associate, Institute of Innovation Research, Hitotsubashi University, Tokyo, JAPAN. 2015-2018 Specialist, SciREX Centre, National Graduate for Policy Studies 2018 Michelin Fellow, CEAFJP/EHESS, Paris, France. 2019- Adjunct Associate Professor, Faculty of Economics, Hitotsubashi University, Tokyo, JAPAN.
Industry/firm level (University/Company) Micro Individual Level (Scientist/Inventor) PATENT - Inventor - Assignee - Patent Number - IPC - Patent Family - Non Patent Literature PAPER - Author - Organization - Category - Acknowledgement DESIGN - No. - Designer Name FUND - No. - Tied Patent/Paper N. Science Linkage Economic Census Innovation Survey(NISTEP) INPUT-OUTPUT TABLE (I/O) Macro Economic Model Funding Database Press Release Survey of Research and Development (Statistics JAPAN) SNA (System of National Accounts; GDP)
Science J-global Data ・# of paper ・# of cited ・Research Categories Convincing three Paper databases to capture scientific activities in global/local journal. Star Scientist Cohort Data Method: Converting XML -> SQL, then Creating Panel Data in the unit of Researcher/Organization (b.)Patent DB PATSTAT (EPO) Patents View(USPTO) J-global (JPO) IIP PatentDB (JPO) Data ・ # of patent ・ # of patent cited ・FI code/IPC code Using three major Patent Office (USPTO, EPO and JPO) to manage Patent Families. Matching DB bet. Patent = Paper Using disambiguation algorism to normalize researcher and his/her organization information. Using Mecab to coordinate Japanese characters (c.)Academic Funding DB SPIAS (SciREX/NISTEP/JST) KAKEN-DB (NII/JST) RePORT (NIH) Nanobank COMMETS (Z&D) Data ・Amount of Fund Budget ・Direct/Indirect Ratio ・Type of Funding Agency Covering Japan (SPIAS, KAKEN- DB) and US (RePORT, Nanobank) Fund data simultaneously (d.) Venture Company Info DB Entrepedia Crunchbase Data ・Carrier/Position of Scientist in Venture Capital Covers Japan (Entrepedia) and US (Crunchbase) database simultaneously, evaluate the economic impact of star scientist Method. Retrieving the data via API, CSV or JSON format. Creating Panel Data in the unit of Researcher/Organization Method. Retrieve the data from Web interface. Method: Converting XML -> SQL, then Making Panel Data in the unit of Researcher/Organization Press Release News Paper
French CNRS と大学の物理学者が対象. 同名のタイトルで公表済み. (Cole and Zuckerman 1984) が明 らかにした, 女性の研究者が男性の研究者に比べてパブリケーションの生産性が低い Productivity Puzzle (gender gap or gender bias) の問題に着目. “The intuitive belief that marriage and motherhood cannot be mesh with a demanding scientific career has been termed an empirically untenable stereotype” (Toren Nina 1991) を踏まえ, 研究のモチベーション は個々人の研究の生産性を測定すること. データの対象は2004年から2005年に活動したフランスの物理学者. 1975年から2005年までのパブ リケーションデータを Web of Knowledge から抽出
EMPIRICAL EVIDENCE AND A PANEL DATA ECONOMETRIC ANALYSIS FOR FRENCH PHYSICISTS AUTHORS; Jacques Mairesse and Michele Pezzoni URL; https://www.cairn-int.info/abstract-E_RECO_661_0065--does-gender-affect.htm Abstracts; PhD Student の論文公刊における, Gender Gap の効果を測定することが目的. アドバイザーの性別の影響, チームの構成や男女比, Graduate Student との関係性などに着目.しラボの構成 が与える影響に着目. 学生の性別, アドバイザーお性別, チームの男女比などを測定するために, STARMETRICS および UMETRICS データベースを活用. ジェンダーの特定には US Social Security レコードを活用. パブリケーションの情報は Web of Knowledge を利用.
EMPIRICAL EVIDENCE AND A PANEL DATA ECONOMETRIC ANALYSIS FOR FRENCH PHYSICISTS (CONT. 2) Abstracts (Cont.); 研究チームの男女構成の影響を測定. チームのサイズは平均8.17. 男性の学生が女性のアドバイザーについた場合はプラスの効果, 女性の学生が男性のアドバイザーについ た場合はマイナスの効果が.
(IBM-GNR), a name search technology produced by IBM the association of names and surnames to one or (more often) several countries of likely origin. Second, the association of names to male and female gender and their associated probabilities. These associations originate from a database produced by the US immigration authorities in the first half of the 1990s, which registered all names and surnames, along with nationality and gender, of all foreign citizens entering the US. It contains a total of roughly 750,000 full names. WIPO worldwide gender-name dictionary (WGND), produced by the World Intellectual Property Organization (WIPO). It includes a list of 6.2 million names from 182 different countries. For each name contained in the dataset, it attaches a given gender to each name by country where that name appears in the original source data
name, IBM-GNR returns the share of instances it identifies as male in the data source and the share it identifies as female. It also returns an additional metric (“frequency”), which indicates the frequency percentile that each name belongs within the complete dataset. 2. For each inventor first name, we attribute female gender to a given inventor if it is identified as female in 97% or more cases and we attribute male gender to a given inventor if it is identified as male in 98% or more cases. 3. When the inventor name is majority female (but not in the 97% of the cases) and the second (or middle) name is 97% or more female, we also attribute female gender. Similarly, when the inventor 5 name is majority male (but not in the 98% of the cases) and the second name is 98% or more male, we attribute male gender. 4. For the remaining 943,725 inventor names, we rely on WIPO’s WGND. As mentioned previously, WGND is a dictionary of 6.2 million names associated with 182 different countries. For each name+country pair, we can match the inventor name to the WGND to determine whether the name has predominantly female or male attribution within that country. To do this, we first need to assign a country of origin to each inventor. (snip) it is able to attribute gender to 3,206,605 inventors, 92.08% of all USPTO inventors
12) For inventors whose surname is primarily associated with China, Singapore, Taiwan, Macao, or Hong Kong (even when they do not reside in those countries), we attribute female gender if it is identified as female in 60% or more cases, and we attribute male gender if it is identified as male in 60% or more cases (threshold decided upon visual inspection of the distribution of GNR’s shares). 13) For inventors whose surname is primarily associated to the Republic of Korea (even when they do not reside in that country), we attribute female gender if it is identified as female in 80% or more cases, and we attribute male gender if it is identified as male in 80% or more cases (threshold decided upon visual inspection of the distribution of GNR’s shares) 14) For inventors whose surname is primarily associated to India (even when they do not reside in that country), we attribute female gender if it is identified as female in 90% or more cases, and we attribute male gender if it is identified as male in 90% or more of the cases (threshold decided upon visual inspection of the distribution of GNR’s shares). After conducting these remaining steps, we attributed gender to 38,188 additional inventors. In total, our (baseline- augmented) method attributes gender to 3,244,813 inventors, 93.18% of all USPTO inventors (6.82% of non-attributed cases)
“Popular Baby Names by Sex and Ethnic Group Data were collected through civil birth registration. Each record represents the ranking of a baby name in the order of frequency. Data can be used to represent the popularity of a name. Caution should be used when assessing the rank of a baby name if the frequency count is close to 10; the ranking may vary year to year.” https://catalog.data.gov/dataset/most-popular-baby- names-by-sex-and-mothers-ethnic-group-new-york-city- 8c742
with a 16-digit number that is compatible with the ISO Standard (ISO 27729), also known as the International Standard Name Identifier (ISNI), e.g. https://orcid.org/0000-0001-2345-6789 Initially ORCID iDs will be randomly assigned by the ORCID Registry from a block of numbers that will not conflict with ISNI-formatted numbers assigned in other ways. ORCID iDs always require all 16 digits of the identifier; they can not be shortened to remove leading zeros if they exist. No information about a person is encoded in the ORCID iD. The identifiers were designed to be usable in situations where personally-identifiable information should/can not be shared. Also, since the ORCID iD is designed to be a career-long identifier, no information that can change over a person's career is embedded in the iD, e.g., country, institution, field of study. https://support.orcid.org/hc/en-us/articles/360006897674
a unique identifier for researchers on Publons. Register on Publons and import your publications from the Web of Science to become eligible for a Web of Science ResearcherID. Each night, Publons assigns a Web of Science ResearcherID to any profiles with one or more Web of Science-indexed publications that do not yet have a ResearcherID. Any publications you add to your Publons profile will then be linked to your Web of Science ResearcherID when anyone searches for you on Web of Science. Please allow up to two weeks for changes you make on Publons to be reflected on Web of Science. https://publons.freshdesk.com/support/solutions/articles/12000038281-what-is- my-web-of-science-researcherid-
Web of Science Core collection. It assigns author ids to the authorships of papers. There are four major components to DAIS Initial Clustering – Starting from scratch, take our whole database without an authority list of known authors, identify the different authors. Ongoing – As new data comes into the database, assign author ids. RID Integration – Integrates manually created publication lists with DAIS Reevaluation – Does a fresh, full clustering on a per name basis; discovers new authors not known at the time of the initial clustering
ID is an author ID that Scopus automatically assigns to each author in its database to group publications of the same author together. For the set of documents grouped under the profile of an author ID, Scopus provides bibliometric information such as citation counts, h-index, and h-graph via its citation overview function. Scopus Author ID is now ORCID compliant. 結局, 複数の Author ID が単一の研究者に紐づけられている可能性がある “Because of author name ambiguity issues and other reasons such as prior affiliations, the automatic matching algorithm of Scopus may generate another new ID for the same author when a new paper is included in the database. ” https://libguides.library.cityu.edu.hk/aim/scopus
our machine to mimic how we need to tell one John Smith from another: run a few search queries. This is particularly feasible because we sit on top of Bing that has indexed many CVs and user homepages that can provide valuable clues. With the entire web at our disposable, we are able to group authors together when doing so will contribute to less than 3% of errors. For more details, please see our January 2018 blog.” 利用している情報 information about author affiliation, publication venues, and co-author network. Our data scientists have developed a method for mining data from authors’ web sites and online CVs. Taking advantage of Microsoft’s web-scale infrastructure, by analyzing billions of documents found on the web, the team has taught the machine to recognize web pages that belong to researchers or may be CVs. https://www.microsoft.com/en-us/research/project/academic/articles/microsoft- academic-uses-knowledge-address-problem-conflation-disambiguation/
Scholar (論文データ) には Author Identifiers を用意 “By default, author values are grouped by their display name, which can result in the aggregation of scholarly works from different authors with the same name. Enabling author identifiers uses the identifiers available in our data sources to group authors, which can help disambiguate different authors with the same name. The author identifiers used currently include Microsoft Academic, and ORCID identifiers if they are available in data from CrossRef or PubMed. N.B. Author disambiguation algorithms can incorrectly assign more than one identifier for the same person. In this case, you may wish to disable this feature to match purely on name alone, or select the different identifiers belonging to an individual author. 実質的には, Microsoft Academics の Author ID を利用 ”
Discussion Paper, 162, http://hdl.handle.net/11035/3215 Hussain, I., Asghar, S. (2017) A survey of author name disambiguation techniques: 2010–2016, The Knowledge Engineering Review, 32, e22. Li, Guan-Cheng & Lai, Ronald & D’Amour, Alexander & Doolin, David M. & Sun, Ye & Torvik, Vetle I. & Yu, Amy Z. & Fleming, Lee. (2014) Disambiguation and co-authorship networks of the U.S. patent inventor database (1975–2010), Research Policy, Elsevier, 43, 6, pp.941-955.
Scientific Productivity ? A Critical Review of the Empirical Evidence and a Panel Data Econometric Analysis for French Physicists, Revue économique, 66, pp.65-113., https://www.cairn.info/revue-economique-2015-1-page-65.htm On-line appendixes to “PROGRESS AND POTENTIAL: A profile of women inventors on U.S. patents”, http://data.patentsview.org/documents/On-line+Appendix+- +Gender+Attribution+of+USPTO+Inventors.pdf
yasushihara
bernhardsvt
signer
aki_moon
nbkouhou
matleenalaakso
japanstrokeassociation
totodo713
cbtlibrary
masakamayama
yuyakurodou
stephaniewalter
sferik
erikaheidi
keathley
chriscoyier
wjessup
geeforr
reverentgeek
kneath
danielanewman
bluesmoon
cromwellryan
![『ギャップを測る』 特許と論文データを用いたイノベーションプロセス におけるジェンダーギャップ分析 2019/12/10 16:30-17:45 於; 広島大学 一橋大学大学院経済学研究科 原泰史 [email protected]](https://files.speakerdeck.com/presentations/30abd9728ecd4f7bb50445373dcde20d/slide_0.jpg){kind=link}
![THANKS [email protected]](https://files.speakerdeck.com/presentations/30abd9728ecd4f7bb50445373dcde20d/slide_70.jpg){kind=link}